Okay I’ll admit it. The returns process for failed hardware isn’t the most fascinating of topics. But it is a part of our job and you’ll need to understand the process. Moreover we expect new engineers to intuitively know about RMAs and DOAs but we never really lay out the process for them.
If you get stuck with an RMA it’s going to suck. I’m documenting this process to help ensure it sucks less. The Network Sherpa blog is here to save you time and hassle. Let’s get to it…
What’s an RMA?
RMA stands for “return material authorization”, and it is permission for you to return a product back to the supplier.
You’ve raised a TAC case and your hardware has been given a grim diagnosis. It’s dead. The TAC engineer will hand you off to the logistics team, who will generate an RMA number for you. The RMA number is distinct from your case number and will require you to log into another portal. Sigh. You’ll be asked to supply the shipping address of the building and a named contact who is on site and can receive the goods. Then they’ll pass on the details to Fedex or DHL to ship to site.
Just beam it to my datacenter
How quickly you receive your replacement part depends on how many suitcases of cash you gave your vendor when renewing your support contract. What a surprise. Your hardware vendor should have depots which are close enough to your datacenter facilities to be able to fulfill the delivery SLA for that failed hardware. This all assumes that your hardware is actually attached to a contract.
You should get the RMA team to confirm the estimated time of arrival (or ETA) for the hardware as soon as possible. Sometimes you’ll assume 4-hour RMA, but later find out that they have you on an NBD contract (details below). Other times the part isn’t available in your local depot (a cache miss). Either way, get an ETA so that you can set expectations with those who will be depending on that part.
RMAs for trade-ins
Note that the RMA process is also used if you’ve cut a deal to trade-in used equipment equipment. [You did know that you can do this?] . When the time comes to ship your old equipment back to the vendor, they’ll use the RMA process to handle the logistics.
So what’s a DOA then?
If your equipment fails before it enters active service, then it’s not an RMA. It is Dead On Arrival (DOA) and as such doesn’t fall under the RMA process.
You could try to raise a TAC case, but you’ll be identified as a DOA and asked to contact the reseller from whom you purchased the equipment. Your brand-new replacement unit will be shipped from the factory as a high-priority. But it won’t come from the support depot, so you won’t get it in four hours! If the part is critical then the misfortune of a DOA could have cost you a week of project time.
Different vendors have different interpretations of what dead on arrival means. Cisco’s guide suggests it that DOA is failure after a few hours. Junipers’ website seems to suggest that a failure within the first 30 days is a DOA, which seems crazy.
Some (ahem) unscrupulous characters have been known to state that a device had already entered service when it failed, so that they received an urgent replacement through an RMA. But remember you may have just paid top-dollar for a used part (see “Where to RMA units come from?” below).
As a cautionary note, be careful who handles your shipping. I once worked for a large carrier who insisted on having vendors deliver to a central depot. Our company would then distribute to each datacenter. Unfortunately my chassis got hit by a forklift and got busted. We had bought direct from our vendor who helpfully said, “sorry lads, it was your shipping company who messed up”. We had to purchase a new router and lodge an insurance claim. Ouch.
Common hardware support levels
2 or 4-hour RMA
This is primo support, and will cost you a pretty penny. Basically the vendor has 2 or 4-hours to get your replacement hardware to site. It’s actually pretty rare to receive the equipment on time unless you’ve done this before. Remember that the clock doesn’t when the hardware fault is diagnosed, but when the RMA number is issued. A missed email requesting your site address can easily cost you an extra hour.
Next Business Day (NBD)
Does what it says on the tin. Even if you have a 2 or 4-hour RMA contract, you can still request the vendor to deliver the next day. You might use this if you have an on-site spare, and wanted to get some extra sleep. Beware though that this is a judgement call. Those few extra hours in bed will be nice, but if there’s a double failure you may regret it!
This is the lowest level of hardware support. Note that the first step after getting an RMA number is for you to ship the failed part to the factory. Then they wait to repair it and send it back to you on a horse-and-cart. Don’t get this level of hardware support unless you’re a massive organization who can manage logistics really well.
But where do RMA units come from?
In many cases these replacement units are used devices which have been repaired or returned to the depot with ‘No-Fault-Found’ (NFF). The depots are frequently restocked with new parts, but you’re only entitled to receive a used part in exchange for your faulty used part.
Given that a vendor recycles hardware between customers, do you really trust them to properly wipe your configuration off the switch before sending it to another customer? Please don’t trust your vendor to wipe your configuration as you could inadvertently leak highly sensitive configuration to your competitors. Not pretty. Develop a policy to wipe your configs before shipping and ensure that it get’s followed.
Sherpas tips for pain-free RMAs
- Provide the name of the on-duty data tech or remote-hands engineer and a contact cell number. Provide these details to the RMA team on first contact. This will increase your chances of a first time delivery.
- Remember that if your replacement part will arrive at or after shift handoff, please add the number of the next oncall datatech on the shipping details also.
- Always get an ETA, and pass it and the shippers tracking number to the people who will be receiving the goods.
- Use your ticketing system to tie your internal ticket number, the TAC SR and the RMA ID together. Some people like to update the title of the ticket with this info so people will know to check the RMA tool.
- The box should hold return shipping labels. If it doesn’t, contact the RMA team. You have 10 business days to get the hardware back to your vendor. After service to the router or switch is restored, you should cut a related ticket to the relevant owner with the RMA details to track the return of the hardware.
- Wipe the startup config before shipping the defective part.
- If you don’t return the hardware in 10 days the ‘collections’ department will come after you for the missing hardware, threatening that they’ll bill you for your now-borked hardware. It’s nice to be able to refer these pests to your tracking ticket and provide the email address of the person responsible for it’s return.